More on Proxy STV (was Re: Capping the Proxy Monster)

Steve Eppley seppley at alumni.caltech.edu
Fri Feb 28 11:07:56 PST 1997


Donald D wrote:
>      The cap must be a function of the number of seats. Steve at
> one time advanced a cap of four percent - which will not
> mathematically fit with any body less than twenty-five.

Donald has misrepresented my position.  I've never advocated *any* 
specific cap %.  I merely mentioned 4% as a possibility after Donald 
had earlier misrepresented me as advocating a 40% cap (a number I 
never even mentioned).

If each member of a 10-seat body were capped at 4%, then each 
would equally have 4% of the weight, and the "missing 60%" would 
be irrelevant since what's relevant is that each member would have
equal weight.  So it's just a matter of semantics whether it makes
"mathematical" sense or not.

But more to the point, it wouldn't "practically" make sense to have
a cap less than or equal to 1/N.  A cap that small would make each
seat equal weight, so it wouldn't be an unequally-weighted Proxy
system anymore.

>      Also you should decide when you are going to transfer the
> excess votes over the cap - it will make a difference in who gets
> elected and also a difference on how many votes each winner will
> have in the lawmaking body. I suggest - here I go suggesting again
> - that the excess votes be transferred at the same time and the
> same way as is done in Preference Voting.

I don't think that's clearly defined, since there's more than one
variation of STV, but, yes, STV is the basic model I have in mind.

Proxy STV would differ subtly from plain STV in how it names winners
(and maybe losers), since there isn't a clear quota needed to win.
Clearly, though, any candidate which polls more than the Hare quota
is guaranteed to be a winner. 

>      When I worked some real ballots from a real election for
> Proxy method, the results had one different candidate elected than
> what I got from Preference Voting. The Proxy winners are:
>    34 A   53 B   46 G   35 H   92 N   32 O   24 P   58 Q   139 S

That's assuming the body still has 9 seats, and that Donald tallied
the votes properly (including weight transfers from eliminated
candidates).  We have no way to check his math, but these results
look plausible.

>      Candidate P is the different winner. I used no cap - the four
> percent cap that Steve suggested would not work on this election.
> If a proper size cap was used candidate P may not have been a
> winner. 

With a "proper size cap" (if there is such a thing) which caps S's
weight, the other proxies would have *more* weight.  Whether P and
others still win, and how much weight they'd earn, would depend on
the transfers from S's ballots, etc.

I request Donald refrain from calling Proxy STV a "monster" until 
he or someone else demonstrates that there would be some important
problem with it.

---Steve     (Steve Eppley    seppley at alumni.caltech.edu)



More information about the Election-Methods mailing list